Skip to content

v1.1

Latest
Compare
Choose a tag to compare
@dayland dayland released this 29 Apr 22:15
· 17 commits to main since this release
878475d

What's New

  • New Chat Modes that include Work + Web, that includes search the Web (via Bing APIs) and compare and contrast answers between my Work documents and Web search results. Also included is Ungrounded which allows direct interaction with an LLM without grounding to allow for creative and fully generative responses.
  • Enhanced File Management for files loaded into IA to including filtering, resubmit and delete file(s). Plus detailed views on processing status and/or error messages without needing IT to look in a database.
  • SharePoint (document libraries, not lists or websites) as a source configurable to point at multiple sites and/or folders. Documents are copied into IA for processing, therefore RBAC is not maintained. The process supports all CRUD operations for files in configured SharePoint document libaries.
  • Compatible with US Gov sovereign clouds including Azure OpenAI hosted in US Gov regions
  • Documentation for use of Bing Safe Search and Azure OpenAI Content Filtering controls for Azure customers to tailor the content filtering behavior to their needs while aiming to prevent potentially harmful generated content and any copyright violations from public content

Preview Features

  • Assistants/ Autonomous Agents are stateful evolution of traditional Generative AI application where you have to manage conversation state, tool integrations, and execute them manually. Assistants/Autonomous Agents are stateful and they automatically manage tool/ function integrations and conversation state. We are introducing two assistants with this release:
    • Math Assistant. How do you get a Large "Language" Model to reason over order of operations in Math? We have included a tool based on Assistants patterns.
    • Tabular Data Assistant. How do you reason over a set of tabular data that you do not know the schema for to answer complex questions in natural language? We show how when a user asks question in natural language and Tabular Data Assistant generates python code to understand tabular data, reason over it, and generate a response.

Coming Soon

  • Secure Deployment: Updates to ensure all traffic and data is private and encrypted. We will also be compatible with Microsoft's Cloud for Sovereignty Security Baselines, Sovereign Landing Zone Baseline for Online and Corp landing zones.
  • Migrate or Upgrade from v1.0 to v1.1 without the need to reload and reprocess all your files.

Issues Fixes

Fixes #295
Fixes #345
Fixes #467
Fixes #481
Fixes #487
Fixes #493
Fixes #496
Fixes #597

What's Changed

New Contributors

Full Changelog: v1.0...v1.1